Introduction

Ecologic studies are investigations in which groups, rather than individuals, are the units of analysis (Morgenstern, 1995). Typically, the groups under study are the resident populations of geopolitical areas such as states, counties, or census tracts; but in principle, studies of other kinds of groups, such as students in certain schools or employees in certain workplaces, would also qualify as ecologic. Studies that involve random assignment of social groups en bloc to alternative conditions also fit our definition of ecologic studies, but for convenience we consider them to be a special kind of randomized trial, as discussed in Chapter 9; here, we consider only observational ecologic studies. Morgenstern (1982, 1995, 1998) has written useful reviews of the ecologic study design.

As an introductory example, Anderson et al. (1998) studied neighborhood environmental factors in relation to the incidence of injuries resulting in hospitalization or death among Hispanic and non-Hispanic white children in Orange Country, California. During 1991–92, injury surveillance was conducted at eight local hospitals and in the county coroner's office. Child injuries by age and ethnicity were counted for each of 594 census blockgroups, which ranged from three to 20 city blocks in size. The estimated number of children at risk and certain socio-demographic characteristics of each blockgroup, including household crowding and the proportion of families with income below the federal poverty level, were obtained from the U.S. census data.